Introduction

Column

Me and my girlfriend with Dance Gavin Dance’s screamer, Jon Mess (Melkweg Amsterdam, 17-03-2023)

Column

Column

This portfolio comprises of a statistical investigation into the genre of Swancore, a “musical hybrid” influenced by progressive post-hardcore, mathcore & Japanese rock, funk & RnB and as of late, mainstream pop. Having done extensive reading, listening and writing in the fields of musicology, ethnomusicology and acoustic ecology, and being particularly invested in the growth of and changes within the Swancore scene, I would like, with this investigation, to combine my interests on a new, statistical frontier. This portfolio is for a course in Computational Musicology at the Universiteit van Amsterdam.

Using data collected by the Spotify API I will address 2 points of interest:

  1. How the push towards a poppier sound (for which I’ll focus on the Spotify API’s “danceability” variable) influences the music’s reception

  2. How this poppier sound informs/is informed by the music’s other qualities (such as Key, Mode, Valence, and Danceability, per the Spotify API).

This investigation centres around 6 major Swancore artists; Dance Gavin Dance (DGD), A Lot Like Birds (ALLB), Royal Coda (RC), Stolas, Body Thief (BT) and Sianvar. This corpus will also include a variety of smaller artists to show how (or rather deduce whether) these central figures have influenced more general trends across the genre as a whole.

DGD’s long lifespan and reputation for having founded the genre places it front and centre in this investigation - in many regards, the changes in sound which this band experienced largely define and inspire the Swancore genre today, and as such I expect their changes in musical qualities will be reflected by other bands. However, as DGD ages and becomes “old news”, RC and other newer additions to the roster of Swancore artists have deviated from the “core” Swancore sound, and in turn these deviations are a better representation of the genre in its current state. Additionally, the contributions to the genre by smaller bands, while inspired by the sounds which came before them, will likely implicate the direction in which the genre is headed in the future.

BT and Stolas should present an interesting parallel, having been formed within a few years of one another and initially possessing a distinctly similar sound, even within Swancore, and having released three albums each, the third of each respectively delving further into mainstream sounds than either band had done before (jazz and funk in Stolas’ case, Western pop in BT’s). But, while Stolas separated in 2018, BT continues to remain together to the present day. I hope that this parallel can help identify how much popularity relies on recency.

As a Swancore-adjacent progressive post-hardcore band, whose discography stretches across a far wider range of influences than DGD or any of the other bands in this corpus, I want to identify if ALLB’s popularity is correlated with its proximity of to Swancore/Poppier influences. Similarily, I would like to take time to identify where Swancore as a genre lies, between its progressive post hardcore roots and pop-funk influences.

Finally, Sianvar is a point of interest as being the first and only (to-date) Swancore “supergroup”, comprising of members of bands throughout the Swancore scene, including Dance Gavin Dance’s Will Swan, Stolas’ Sergio Medina, A Lot Like Birds’ (and later Royal Coda’s) Joe Arrington and Micheal Franzino, and Hail the Sun’s Donovan Melero. While the band is currently on hiatus, it should serve as an interesting look into a very specific point in Swancore’s history - right at the cusp between second-generation Swancore and the Swancore we have today.

One of the primary issues with assessing such a niche and complex genre is how data about it is gathered and catagorised. As the Spotify API runs on relationality (placing all data points in relation to one another), a genre still so linked with the less accessible (both musically and socially) post-hardcore scene will not have as large a range of popularity, danceability, and other subjective characteristics as more mainstream music. As such, this portfolio will focus more on empirical qualitative trends than quantitative ones.

Tracks of interest in this corpus will be DGD’s Summertime Gladness, ALLB’s Recluse, BT’s Death, Drugs & Dancing, Stolas’ Medusa (feat. Kurt Travis, Sianvar’s Sick Machine and from smaller artists, Dwellings’ Pick Up Before You Go.

This little introduction aside, I hope you click, scroll, read and listen on, and that you enjoy the output of these past few weeks!

Column

Swancore: an Overview

Column

Analysis

Before getting to the more pressing questions of my corpus, I want to give a brief overview of it (and consequently Swancore as a genre) in its entirety. As can be seen in the following graphs, Swancore typically has an incredibly high level of Energy (averaging ~0.9 out of 1) despite a relatively low average Valence, or “happiness value” of ~0.4 out of 1. This is likely because, as seen in the Mode graph, the more common mode of Swancore music is minor. These may initially seem like contradictory factors, but actually align with Swancore’s identity as having found its roots in the Progressive Post-Hardcore scene, which is generally known for its eclectic mix of quick tempos (see the large concentration of Swancore tracks with a Tempo over 150bpm), complex riffs and time signatures and eccentric screaming with mellower clean vocals and instrumental segments, and its tendency to tackle emotionally heavy concepts.

While the Key and Time Signature bar graphs provide some interesting knowledge, indicating the relative lateralness of Key preference (with the strange outlier of D# keys) and an overwhelming preference for 4/4 time, they are not particularly relevant for this investigation other than in discovering inaccuracies within the Spotify API: through interviews with various bands in this corpus, the time signatures of various songs have been revealed to be far more complex than Spotify makes them out to be. For instance, Spotify failed to recognize any songs in 7/4 time, despite there being multiple in the playlist, such as A Lot Like Birds’ Orange Time Machines Care (revelaed by drummer Joe Arrington in a Rocksound interview). This bears consideration when discussing the danceability of tracks, because if the API can not recognise a time signature accurately, doubts arise about the API’s capacity to gauge how well someone might dance (or not dance) along to a track.

Most importantly, this overview provides the baseline values for Danceability and Popularity for the rest of this investigation: the average danceability of the entire corpus is ~0.4, while the average popularity is 29. The hypothesis I would like to propose is that, as Swancore grows as a genre, its Danceability and Popularity will rise along with it (and consequently far overshoot both of these averages) as it steadily incorperates more elements of pop (particularly within tempo and structural characteristics).

Column

Energy

Valence

Mode

Tempo

Time Signature

Key

Popularity

Danceability

The Push to Pop: A Timbre Analysis

Column

Timbre Coefficient Comparisson

Column

Analysis

Before we get to investigating the danceability of Swancore, how can we be sure that it is deviating from its Progressive Post-Hardcore roots? And how can we be sure that it’s heading towards Pop? This violin graph (which depicts the range and frequency of the intensity of timbre coefficients of groups of songs) should help represent the place which Swancore holds relative to other genres which are of sonic proximity to it: Progressive Post-Hardcore, Funk Rock, K-pop and Pop.

What is made clear from this representation is that Swancore situates an In-Between of Progressive Post-Hardcore and Funk Rock, while bearing some closer resemblances to Pop music. These similarities are particularly visable in timbre coefficients 02, 04, 07, 08 and 012, which represent various aspects of “brightness”, “thickness” and “heaviness” of pieces, respectively (If that sounds vague, that’s because timbre is too! It makes very little sense to me to quantify timbre, particularily when it seems exclusively qualitative, and yet, here we are!)

Dance Gavin Dance

Column

Dance Gavin Dance’s Discography

Column

Analysis

This plot (and all the others in this “Discographies” tab) depicts the relationship between danceability, popularity, and release date across the entire discography of the band (or bands) in question.

Immediately upon viewing this graph it becomes clear that Dance Gavin Dance’s discography features an upward trend in danceability starting from 2013. This trend makes sense; this year contained the first release with their newest and current vocalist, Tilian Pearson, whose influences notably lie in the genre of alternative pop, as visible in his solo career. The trend’s gradual accumulation checks out as well - when he first joined DGD, Tilian was fresh off the heels of his time singing for Tides of Man, an alternative rock band with a grittier, heavier, and discernibly less poppy sound. As his time with DGD progresses, Tilian’s influences have become more predominant in the band’s sound.

Through this graph we are also able to identify some outliers - notably the band’s most and least danceable songs, and an entire album which, for having a considerably pop- and funk-influenced sound, is rated on average as not being particularly “danceable”. These, respectively are 12 Hours, 630 Miles in the top left, Untitled 2 in the bottom right - both of which are interlude tracks with limited vocals - and the erratic, high-energy 2011 album Downtown Battle Mountain II.

Interpreting these outliers can provide some insight into how Spotify’s API measures danceability. In seeing that 12 Hours, 630 Miles, the slow outro to their 2007 album Downtown Battle Mountain, which has a consistent tempo, rhythm and very little variation in timbre throughout is considered incredibly danceable, while Untitled 2, the introduction song to 2023’s Jackpot Juicer continuously builds up (i.e. changes timbre) to a climax that occurs in the following song, Cream of the Crop, one can hypothesise that consistency is a central factor in Spotify’s calculations of a given track’s danceability. The relatively low placement of tracks from Downtown Battle Mountain II would coincide with this inference, as the musical complexity (and at times incoherency) of the 2011 album far exceeds anything else the band has released, and could overrule their funk-influenced timbre in the Spotify API’s calculations.

A Lot Like Birds

Column

A Lot Like Birds’ Discography

Column

Analysis

When interpreting A Lot Like Bird’s graph, it is important to understand that the band, which is more influenced by (and thereby representative of) the progressive post-hardcore genre than it is of Swancore - with this knowledge in mind, it makes sense that its trend is a negative one, with danceability decreasing as time progresses and the band delved into a more eclectic and experimental style. Similarily, the peak in 2011 with the release of Conversation Piece aligns with Kurt Travis (DGD’s former clean vocalist) joining the band, and bringing his funk influences with him.

Lo and behold, yet again two outliers and “most danceable” tracks in a band’s discography are interludes or non-songs: Abbr. from Conversation Piece is a track which does not feature any variation - if anything, its consistency is its defining feature in an album which is otherwise fairly erratic. Similarily, In Trances, the introduction piece to 2013’s No Place is a spoken word poem supported a repeating sample of a grandfather clock chime over a trap beat.

On the other end of the scale is Recluse from No Place. According to drummer Joseph Arrington on Rocksound “If there was ever a song that combined ideas, styles, chaos, melody and batty time signatures…this song is even worse (better?!) We took our strange phrasing ideas to the absolute limits with this song”. This description of the song’s “madness” is more than apt, its polyrhythmic foundation of 4/4 kick and 6/8 snare patterns are likely the reason for this song’s lack of dancability. Like with Dance Gavin Dance, ALLB’s characteristics point towards Spotify API focusing on consistency when deriving a track’s danceability.

Royal Coda

Column

Royal Coda’s Discography

Column

Analysis

A number of points of interest arise from this graph. Firstly, and most noticeably, the spike in danceability between Royal Coda’s releases of their 2018 and 2019 albums, Royal Coda and Compassion, respectively. This shift aligns with the exchange of the band’s supporting guitarist, Thomas Erak (formerly of post-hardcore band Chiodos and currently of mathcore band The Fall Of Troy) for Will Swan (of Dance Gavin Dance), who brought along with him the pop and funk influences which Dance Gavin Dance was going through at the time. Here it is also worth noting that RC’s lead vocalist, Kurt Travis was Dance Gavin Dance’s second vocalist, and was one of the main contributors to the band’s initial adoption of funk influences - hence it is likely that reuniting musically with one of his former bandmates could have inspired the change in sound between RC’s first and second albums.

Secondly, with regards to noticeable outliers, three that stand out: See Them Faceless from the self-titled Royal Coda album, All In Question from Compassion, and As We Fall Into Deep Waters from 2022’s To Only A Few At First. In line with the hypothesis that Spotify calculates danceability by consistency, See Them Faceless’s consistant tempo and relatively uniform dynamics (volume) make for what the API considers a more “danceable” experience. On the other hand, All in Question’s wide changes in dynamics, tempo, and even time signature, and As We Fall Into Deep Waters’ shifts between faster and slower segments make them less danceable.

Finally, taking a look at the two interlude tracks across RC’s discography, Suffolk and The Innocence Of - Reprise (from Royal Coda and Compassion, respectively), these tracks also reflect the danceability hypothesis: Both initially sound like they’d have an even danceability, featuring an acoustic riff supported by quiet. lyricless choral vocals, but Suffolk transitions from a 30-second sample of classical music into its acoustic riff, then into a loudening sound of static, whereas The Innocence Of - Reprise only consists of its main section. This likely shows that quantitative inconsistency in sonic qualities (whether of tempo, timbre or volume) take precedence in calculating danceability, their qualitative characteristics (such as a human understanding that quiet volumes and slow tempos being less enticing of dancing).

Stolas

Column

Stolas’ Discography

Column

Analysis

From first glance, and contrary to the perception that Swancore (following Dance Gavin Dance) is moving towards a more danceable sound, Stolas’ chart seems to indicate the opposite, that the band reached a peak in danceability with its second release, 2014’s Allomaternal, and leveled out for its final release, the self-titled album, 2017’s Stolas.

However, it is worth noting that for the band’s two earlier albums, the majoraty of songs are less danceable than from their third, self-titled one, and that a number of outliers do heavily influence the trend of the graph: Thief & the Hourglass and Captured Light from 2013’s Living Creatures, Somatic from Allomaternal, and Euphoria from Stolas.

Of these, Somatic stands out as particularly clear support for the danceability hypothesis being investigated: the 5:11 minute song features a constant tempo and dynamics, and most importantly, a distinct rhythm. Whereas other songs on Allomaternal are truer to Stolas’ eccentric and erratic sound, Somatic’s consistency and trance-inducing repetition (which align with the song’s lyrical context - being a dream of the album’s protagonist according to bassist RJ Reynolds) contribute to the Spotify API’s interpretation of a relatively relaxed song as being a danceable one, and ultimately increases the average danceability of Allomaternal as an entire album.

This being said, there is an alternative interpretation of the shape of Stolas’ graph: a concept named Recapitulation. A phenomenon I have previously approached in Returning to the Tonic: Using Swancore to Expose Aesthetic Taxonomies as Restrictive Historical Agents in Music, the conceptual consolidation of a genre’s “ideal” sound is often responsible for inspiring gradual homogenisation in its sound over time (A cyclic process I denoted as “recapitulation”). Arguably the first band formed in the second generation of Swancore (ca. 2012 - 2017), Stolas’ sound is still largely influenced by external genres, rather than being influenced by a perceived notion of what Swancore “needs” to sound like. In this manner, Stolas’ diversion from the trends of Dance Gavin Dance and other Swancore bands could also be a result of having maintained a higher degree of creative input (conscious or otherwise) over their sound than the genre within which they are trying to write. Indeed, while newer Swancore bands tend to follow this trend towards danceability and pop, Stolas notably took influence from avante-guarde scenes when writing music for Allomaternal and Stolas, allowing them to secure for themselves a truly unique and (in my opinion) under-appreciated niche within the Swancore genre.

Body Thief

Column

Body Thief’s Discography

Column

Analysis

For initially having a sound closely related to Stolas, Body Thief’s graph tells a much different story: whereas the former’s danceability tapered off with successive releases, the latter’s has one of the most distinctive increases over time, across every band in this corpus. This difference quite possibly correlates with the bands’ relatively different lifespans: Stolas was formed many years earlier and released its first album 2 years before Body Thief, and ended its career in 2018 (though they will reunite in June of 2023 for one night for the anniversary of Living Creatures’ release). Body Thief’s relative longevity, surviving 4 years longer, and recency via persisting to present-day, has allowed the band to adapt to newer trends in the Swancore sound.

Shame as it is, familiarity and consistency - two of the main characteristics of pop (and most popular music for that matter) - are fundamental to its success, and tend to . These characteristics are most prominent in Body Thief’s latest album, 2022’s Every Ending, with songs like Death, Drugs and Dancing, Exit Wound All Our Love featuring the familiar pop “verse -> chorus -> verse -> chorus -> bridge -> chorus -> outro” structure, a major deviation from their previous albums more free-flowing organisation of verses and choruses, such as in Night Owls in Mating Season from 2019’s Travel Glow. It comes as no surprise, then, that when looking at the graph, the 5 most danceable tracks all come from Every Ending (An album all whose tracks are above a 0.4 danceability, with most over 0.5), while the 5 least danceable (even excluding the to-be-mentioned outlier) are all from their earlier albums, with a majority from Speak in Hibernation.

In terms of outliers, an outstanding one is Parasites in the Water from 2015’s Speak in Hibernation. This track has something of a Suffolk (from Royal Coda) Syndrome, whereby the track enters and exits with a sample of a completely different tempo, timbre, and so forth. In this song’s case, completely different to Suffolk it is actually a sample of unidentifiable tempo as it is of, per the song’s name, the sound of running water and the lapping of waves. This disruption of Spotify’s tempo analysis certainly contributes to the Parasites in the Water’s incredibly low danceability (within the bottom 5 of this entire corpus), but also unlike Suffolk, the song is not particularly consistent to begin with - this would explain the difference of nearly 1 full unit of danceability seperating them.

Sianvar

Column

Sianvar’s Discography

Column

Analysis

Given that there is not much data in this plot of Sianvar’s discography, it is hard to make any guesses about the direction in which the band would have gone had it continued to exist. However, that is not to say that this data is useless: it serves as an apt snapshot of contemporary Swancore’s transition from a more chaotic, varied genre into a more consistent sound: seen in the change from the first album, Sianvar which has both the band’s third most danceable song, Your Tongue Ties and least danceable, Substance Sequence (which also ties for second-least danceable in the entire corpus), to the much more clustered second album, Stay Lost, whose tracks all fall in a small range of danceability which is higher on average than the first album.

However, the danceability of Your Tongue Ties throws a bit of a cog in the works of the hypothesis that danceability relates to consistency of tempo. While verse and chorus tempos are consistent with one another, the intro and post-choruses send the piece into an entirely different tempo and time signature, which should disrupt the Spotify API’s capacity to detect a common tempo. Perhaps the length of the piece solves this issue in a way that isn’t possible for Suffolk and Parasites in the Water (Royal Coda and Body Thief, respectively), but this is only speculation.

Smaller Artists

Column

Smaller Artists’ Discographies

Column

Analysis

Needless to say, this graph comes as a bit of a surprise. With relatively little change in the average danceability across nearly a decade of releases, one is led to believe that the genre as a whole has not experienced a change in sound, contradicting both my earlier claim and the trends shown in larger bands’ discographies. So what’s at play here? Largely, it seems to come down to individual outlier songs, and the influence of a few key bands. Before I describe these, I would like to apologise: I tried for the life of me to find a way to show the name of each piece’s artist, but could not figure out how to do so. It is unpleasant, but you can double-click on the album names in the legend to isolate them from the rest of the graph, and this should allow you to isolate each of the individual smaller artists’ discographies for analysis.

First, looking at outliers of incredibly high or low danceability, a few are immediately evident: Interlude from Oranges’ album Taxonomy on the high end, and I Met A Yeti’s Honey Bee (from Camp Yeti) and Anemoria’s single, Lunatic Fringe, on the low end, respectively. As implied by its name, Interlude is an interlude in a larger album, and likely has a high danceability due to the reasons discussed with regards to 12 Hours, 630 Miles and Abbr., namely a short duration, consistent tempo, rhythm and time signature. With regards to Honey Bee and Lunatic Fringe, both feature incredible variations while also maintaining complicated rhythms in instrumentals and vocals alike throughout.

Working with Lunatic Fringe in particular, and looking at Anemoria’s work in general, their employ of funk influences in timbre and phrasing has not altered the chaotic structure which is typical of early Swancore and Progressive Post Hardcore, and as such most of their releases (Anemoria - EP, Cashmere Jungle, Jellybeans, Asomatous, and Lunatic Fringe) are below the average danceability of the Smaller Artists corpus. Alongside Anemoria are I Met A Yeti (Camp Yeti, Red-Eyes B. Yeti, Cthuwu) and Adam in the Bayou (Bon Ami), both of whom seem to emulate the sounds of early Stolas and Dance Gavin Dance, respectively. Coincidentally, I Met A Yeti’s singer, Daisy Chamberlain, also features on the not-very-danceable Resilia’s Royal Flush, though that song’s lack of danceability likely arises from the unpredictable switches between regular and irregular bass-and-snare drum rhythms and fills. Joining their ranks seems to be AnimalJam (AnimalJam, The Process of Dissention, Anhedonia), who have had a downward trend in danceability across their discography thus far.

On the other hand, artists within the Swancore scene such as Dwellings (Lavender Town, Pick Up Before You Go, Devices) and nightlife (new low, fallback) seem to be contributing to an even more structured, groovy (and more importantly, danceable) sound for the genre.

Ultimately, it seems where danceability is involved, that there is a divergence in Swancore. Some bands are tending towards the more chaotic sound of the genre’s roots, while others are more inclined to explore a more structured, mainstream approach to the hybrid fusion.

Dance Gavin Dance’s Summertime Gladness

Column

Chromagram Summertime Gladness (Dance Gavin Dance, 2017)

Column

Analysis

This page hosts two graphical representations of Summertime Gladness, one of Dance Gavin Dance’s most popular singles, and their most danceable according to Spotify’s API.

The Chromagram, which represents the frequency of specific pitches being played in a piece of music, registers large sections where the pitches being played are overwhelmingly G# and F#, and others where A#, D# and C# are more prominent: this difference outlines the seperation of verse and pre-chorus segments, which are chords and arpeggiated riffs involving those notes, respectiely. The bright yellow section of F# at the end is likely because of the shrill, sustained saxophone notes which play during the song’s outro.

The Cepstogram, which represents the intensity of specific timbre coefficients across a song’s duration, indicates that the song is incredibly bright, based on the intensity of timbre coefficient 02. At the transitions from verse to pre-chorus (such as at 0:48), the change in timbre coefficient 05 (or thickness of the music) coincides with the reduction in number of instruments playing (removing the drums, bass, screamer and rhythm guitar from the section). Timbre coefficient 03 seems to match quite closely with the onset of each lyric/word, which leads me to believe that it is picking up on a certain vocal quality of each of the singers (perhaps smoothness/roughness as to match with the differences between Tilian’s clean vocal sections and Jon’s screamed ones).

Column

Cepstogram Summertime Gladness (Dance Gavin Dance, 2017)

Sianvar’s Sick Machine

Column

Chromagram Sick Machine (Sianvar, 2014)

Column

Analysis

This page hosts two graphical representations of Sick Machine, Sianvar’s second least danceable and popular song, according to Spotify’s API. It also happens to be one of my favourites… so everybody else must be wrong ;)

The chromagram shows a prolific presence of the pitch B throughout the entire piece, particularly during the first 30 seconds of the song: this links with the distorted guitar drone which backs up the drum fills serving as an introduction. the following erratic representation of notes coincides with the actual piece, which plays a repeating riff of deafeningly fast ascending steps, which constitutes much of the remaining piece excluding the nearly minute-and-a-half long outro which consists of a similar prolonged set of the pitch E, travelling occasionally to a powerful A, G and B before ending the piece.

The cepstogram features a very intense coefficient 02, which is associated with the piece’s brightness. This makes sense, both the tone of guitar and voice being quite light (though with distortion, almost nauseating), but is also quite ironic considering the heavy lyrical content of the song, which viscerally covers the position of soldiers in a combat, while making pointed references to the Vietnam war. Additionally, my proposition that coefficient 03 deals with smoothness or roughness of voice seems to be correct, as at 1:01 singer Donovan Melero’s softest section occurs, and at 1:20, his roughest scream. Both timestamps have the most intense moments of c03 in the song.

Column

Cepstogram Sick Machine (Sianvar, 2014)

Body Thief’s Death, Drugs & Dancing

Column

Timbre Self-Similarity Matrix

Chroma Self-Similarity Matrix

Column

Analysis

Death, Drugs & Dancing, from the album Every Ending, is one of Body Thief’s most popular songs, and their most danceable. The consistency which has hitherto been associated with danceability continues to prove accurate, as the song’s timbre self-similarity matrix does not have many lines of intensity indicating novelty (with the exception of the piece’s bridge, at approx 160 seconds). The dark checkerboard pattern which is present is indicative of repetition.

Looking at the chroma self-similarity matrix, the lack of secondary or tertiary diagonal lines parallel to the one along x=y indicates that there is enough variation to keep it from being direct repetition, but the grain of the graph also shows that the actual pitches themselves are repeated quite frequently.

Stolas’ Medusa (feat. Kurt Travis)

Column

Timbre Self-Similarity Matrix

Chroma Self-Similarity Matrix

Column

Analysis

Completely unlike Body Thief’s timbre self-similarity matrix, Stolas’ Medusa (feat.Kurt Travis)’s, from their album Living Creatures, is chalk-full of novelty from beginning to end. The first intense change is at 26 seconds, with the shift from a distored ebb-and-flow to a powerful guitar riff, and then again with the introduction of drums and rhythm guitar at 0:31, and then yet again with the introduction of screamed vocals at 0:51. The song continues at the same pace for a while, until around 1:30, where a bass version of the intro riff takes hold of the piece. Then; Kurt Travis enters, throwing the entire piece on its head with a slow, soothing melody. This persists, and mellows out, most of the novelty of the piece explored until 2:50, with the introduction of yet another verse structure to the piece, followed by another new introduction of a guitar solo. The outro, removing the drums entirely for a soft vocal exit adds the final spin to the piece.

Looking at the chroma self-similarity matrix, the chaos of the piece is truly captured by the vibrant array of pitches being represented at all most times - a distinct exception being for the aforementioned Kurt Travis feature, in which the intensity pulls back so that his voice receives the most attention for that section. The other section with relatively low chroma representation is the introductory ebb-and-flow guitar section, which is only one note. The rest of the song, however, epitomises the chaos of the early Swancore sound.

It’s no wonder that this is the second least danceable song in Stolas’ repertoire, even if it’s one of the coolest.

A Lot Like Birds’ Recluse

Column

Tempogram: Recluse (A Lot Like Birds, 2013)

Column

Analysis

As discussed in the section on A Lot Like Birds’s discography, the chaotic and “batty” (per Joe Arrington) Recluse from the album No Place is the band’s least danceable song. This tempogram, a visual representation of the changes in tempo across the song’s duration, should help gauge the accuracy of my earlier hypothesis regarding how Spotify’s API conducts its measurement of a given track’s danceability.

Represented by lines of concentrated yellow, Recluse’s tempo is shown to modulate quite dramatically at many parts of the track, with intense and fast-paced sections followed by much slower and drawn-out ones. Additionally, there are points in the track where the analysis is not very distinct or readable (such as between approx. 60 and 100 seconds into the track, where there are hardly any horizontal, “even” tempos represented by yellow lines, but rather an array of vertially oriented yellow dots projected seemingly at random). This inaccuracy is likely due to the difficult polyrhythmic drum patterns on which the piece is based, which I discussed in the earlier section (as a review, Arrington plays a 4/4 kick drum underneath a 6/8 snare pattern, each part of which guides a different section of the other instruments - Bass and the band’s screamer follow the kick drum, while the guitars and clean vocalist follow the snare). Another possibility is the software confusing the heavy distortion throughout the song for percussive presence, which would confuse the tempo of the piece.

This would explain why, between approx. 125 and 135 seconds, where there is a reduction in the thickness of the song’s sound (guitars, bass and clean vocalist pull back, allowing drums and harsh vocals a much clearer presence), the graph is at its clearest, depicting a slight accelerando, or increase in speed, around 250bmp, along with some less distinct but still visible tempo harmonics at 400 and 500bmp, respectively.

In the end, it is likely that this song’s structural complexity, in conjunction with strong variation througout the piece and an imprecise analysis by the Spotify API, are what contribute foremost to its low danceability. Expanding from this, it would explain why a good number of songs from ALLB’s (and other Swancore bands’) discographies are considered to have a low danceability despite having a discernably poppier timbre and influence: their structural complexity and variation impede upon their percieved ability to make someone dance. You can have an incredibly funky riff (such as in Vanity’s Fair from ALLB’s Conversation Piece), but it on its own won’t make the piece more danceable if it’s surrounded by grittier, inconsistent segments.

Dwellings’ Pick Up Before You Go

Column

Pick Up Before You Go (Dwellings, 2019)

Column

Analysis

Dwellings, a band known quite well within the Swancore community for having given the genre a soul of its own (in both the figurative and genre sense of the word), released Pick Up Before You Go as a single in 2019. The song, which is 14th most danceable in this entire corpus, is exemplary of the “consistent structure = danceable” hypothesis. The tempogram, which shows a clear tempo and tempo harmonic steadily throughout the piece, breaking only for a brief ritardando at 2:24 where the piece has its bridge. Otherwise, the tempo is unchanging throughout.

The perceived tempo of approx 240bpm seems incredibly high (because it is), but on closer listening it appears that the Spotify API has latched onto the guitar and synth 16th notes which play in the intro and verses as the basis for tempo as opposed to the drums underneath. In this case, while still officially incorrect, the Spotify API still has the correct relative tempo as it is measuring using a tempo harmonic.

Dendrogram and Heatmap

Column

Dendrogram: Clustering Similar Swancore Songs

Heatmap: Identifying Clustering Processes

Column

Analysis

Working with a reduced playlist consisting of some of the most danceable songs from each album in the corpus, this cluster dendrogram attempts to cluster groups of songs together based on similar qualities. As can be seen in the heatmap tab, these traits are the timbre coefficients, energy, valence, acousticness, speechiness and loudness variables provided by the Spotify API.

Looking at which songs cluster together, a few things stand out: primarily, the clustering of songs with the same singer, or same instrumentalists. On the left of the dendrogram, four songs of Dance Gavin Dance’s where Tilian is performing clean vocals (Pray To God For Your Mother, Suspended in This Disaster, Summertime Gladness and Calentamiento GLobal) have been clustered together, broken only by two Sianvar tracks featuring Will Swan as their guitarist (Coordinate Love, and Your Tongue Ties).

However, the further up you go, some more interesting connections are made: for instance, the two tracks last incorperated into the cluster both feature Kurt Travis on vocals. In fact, for a singer whose songs make up nearly a 3rd of this dendrogram (7 songs in total across 3 bands), his songs are relatively spread out across the dendrogram. Perhaps this indicates that he is an incredibly versatile singer, or that he continues to find ways to make his same funk inspirations sound new within the genre he has learned to call his own.

In the heatmap, we are better able to see similarities between different subjectivity variables (such as the timbre coefficients) and the songs which they are trying to assess. On the heatmap, sonic proximity is denoted through a similarity in colour. For instance, for many songs, and Uneasy Hearts Weigh The Most in particular, timbre coefficient 06 and acousticness have a deep-seeded similarity (shown in their similar shade of bright yellow), which leads to them clustering with one another in the first stage of the heatmap.

This is an interactive graph, and really quite intuitive, so I suggest listening to these songs individually and figuring out for yourself whether you agree with the way in which they match up.

Conclusion

In conclusion, we were able to discern the answers to the two questions asked in the introduction:

  1. Yes, with an increase of danceability, popularity too quickly follows: Looking no further than the work of Body Thief and Dance Gavin Dance on their most recent albums does this become clear.

  2. The push towards danceability less so influenced the music’s other features, than was influenced by them - with more repetitive and coherent songs being considered more danceable by the Spotify API. However, this API too has some issues, which we came to realise:

We’ve discovered too, that the Spotify API has a hard time discerning whether a track is actually meant to be danced to: Interludes, Reprises and Abbreviations of all musical sorts were among the most danceable in the corpus, despite often being placed where they were to offer a breath of fresh air from all the head-banging these albums are meant to induce. Perhaps, if the API were to consider track duration into the danceability variable, it would be more accurate with circumventing these issues. Or, perhaps it doesn’t care: after all, danceability has nothing to do with intent, only capability. Regardless, it’s been a fun (if stressful) project, and it is time to lay down my computer.

Thank you for coming on this statistical journey with me. It was fun.

I do not think I will be doing it again.

Cheers, Richard Weichert